The received wisdom is that there isn’t much difference between charter school performance and the performance of other public schools, with some of the newer evidence being that certain kinds of charters, those following the “no-excuses” model do indeed get better results. Most of the evidence we have looks at test scores rather than longer-term outcomes.
In contrast, Tim Sass and colleagues have taken a look at longer-run outcomes for charter schools in Florida and in Chicago. Their evidence shows that students who attend charter schools are more about 10 percent more likely to start college and stay in college than are other students. They also find earnings a decade later are higher on the order of $2,000 to $4,000 per year. These are rather large effects.
First, a direct look at a few of the authors’ results. Then a discussion of some of their caveats.
These are some of the educational outcome results. Note particularly the results on “persisting in college.” As the authors point out, many students start college but drop out within a year or so. So their persistence measure is a good way of checking for real effects. The Florida estimates are large. The Chicago estimates are not quite so large, and the smaller sample means we can’t be sure that the results are statistically meaningful. Overall, the results make charter schools look pretty good.
Of course, the proof of economic impact is to look at earnings in adulthood. Here’s a snippet from the paper just for Florida.
The problem with these kinds of studies is always separating out the effects of going to a charter school from the possibility that students who go to charter schools were, in some way, more talented to begin with. The authors control for this possibility in two ways. First off, they include control variables for observable student characteristics. That’s good, but it leaves open the possibility that there are unobservable differences between charter school students and others.
The clever idea the authors use is to compare students who were in a charter school in 8th grade and then moved to a noncharter school to students who remained in a charter school through high school. The idea is that we know that whatever unobservables lead to choosing a charter school are alike in the two groups, as everyone in the sample did indeed choose a charter school.
The technique isn’t perfect, as we don’t know whether the difference on staying in a charter school is random or indicates some important difference in student ability. Nonetheless, the authors have nicely pushed part way toward getting rid of the “selection on unobservables” problem.